Apical splenic nerve electrical stimulation discloses an anti-inflammatory pathway relying on adrenergic and nicotinic receptors in myeloid cells

The autonomic nervous system innervates all lymphoid tissues including the spleen therefore providing a link between the central nervous system and the immune system. The only known mechanism of neural inhibition of inflammation in the spleen relies on the production of norepinephrine by splenic catecholaminergic fibers which binds to β2-adrenergic receptors (β 2-ARs) of CD4+ T cells. These CD4+ T cells trigger the release of acetylcholine that inhibits the secretion of inflammatory cytokines by macrophages through α7 nicotinic acetylcholine receptor (α7nAchRs) signaling. While the vagal anti-inflammatory pathway has been extensively studied in rodents, it remains to be determined whether it coexists with other neural pathways. Here, we have found that three nerve branches project to the spleen in mice. While two of these nerves are associated with an artery and contain catecholaminergic fibers, the third is located at the apex of the spleen and contain both catecholaminergic and cholinergic fibers. We found that electrical stimulation of the apical nerve, but not the arterial nerves, inhibited inflammation independently of lymphocytes. In striking contrast to the anti-inflammatory pathway mechanism described so far, we also found that the inhibition of inflammation by apical nerve electrical stimulation relied on signaling by both β 2-ARs and α7nAchRs in myeloid cells, with these two signaling pathways acting in parallel. Most importantly, apical splenic nerve electrical stimulation mitigated clinical symptoms in a mouse model of rheumatoid arthritis further providing the proof-of-concept that such an approach could be beneficial in patients with Immune-mediated inflammatory diseases.


Introduction
The autonomic nervous system innervates lymphoid tissues, therefore providing an important link between the central nervous system and the immune system. While previous studies have revealed direct autonomic innervation of parenchymal tissue in the thymus, bone marrow, spleen, lymph nodes, and gut-associated lymphoid tissues (Nance and Sanders, 2007), the anatomical and functional characterization of the nerves that project to these lymphoid organs remain incomplete from an anatomical and functional point of view. This is the case for the spleen which is the largest secondary lymphoid organ in the body and the main source of pro-inflammatory cytokines in systemic inflammatory diseases (Murray and Reardon, 2018).
Studies performed in the seventies have shown that the splenic nerve in humans carries approximately 98% sympathetic nerve fibers (Heusermann and Stutte, 1977;Kudoh et al., 1979). Catecholaminergic postganglionic nerves originating mainly in the superior mesenteric/ celiac ganglion enter the spleen accompanying the splenic artery and run along the trabeculae in plexuses. Nerve fibres from the vascular and trabecular plexuses enter the white pulp along the central artery, where they reach their greatest density and end up in the periarterial lymphatic sheath. Sympathetic nerve fibres are co-localized with T-cells, macrophages, as well as B-cells residing in the marginal zone where lymphocytes enter the spleen (Anagnostou, 2007;Hoover et al., 2017). Catecholaminergic innervation is particularly rich in T-cell zones and in areas of mast cells and macrophages, whereas follicular and nodular zones where B cells mature, are poorly innervated. In agreement with what has been observed in humans, Felten et al. have (Felten et al., 1987). Experimental studies in rodents have shown that sympathetic nerve terminals in the spleen are able to store and release norepinephrine in response to stimulation (Elenkov and Vizi, 1991;Kees et al., 2003), and that the splenic norepinephrine content dramatically decreased following chemical (Sudo, 1985) or surgical  sympathectomy.
Further experiment from the Tracey's group demonstrated that the splenic nerve was a critical component of the vagal anti-inflammatory pathway, a physiological regulatory mechanism whereby afferent vagus nerve stimulation by pathogen-derived products leads to efferent vagus nerve-mediated suppression of proinflammatory cytokine production by spleen macrophages in the red pulp and the marginal zone (Rosas-Ballina, 2008). This was elegantly demonstrated by the surgical ablation of the splenic nerve which abolished the inhibition of LPS-induced TNF production by vagus nerve stimulation in mice. The vagal antiinflammatory pathway relies on the production of norepinephrine by Fig. 1. Anatomical, histological and functional characterization of murine splenic nerves. (a) Location of splenic arteries in an anaesthetized mouse using laser speckle contrast imaging. White arrows indicate the location of nerve-like structures. (b) Representative microscopy imaging of the indicated nerves stained for Tyrosine Hydroxylase (TH, red), Choline Acetyl Transferase (ChAT, green) and Neuro Filament (NF, blue). Scale bar = 30 µm. (c, d) Light sheet imaging of whole spleen (c) and apical region (d) after staining for TH (black (c); red (d)) or ChAT (black (c); green (d)). Red and green arrows indicate points of entry of arterial and apical splenic nerves, respectively (c). Scale bar = 1 mm (d). (e, f) Arterial or apical splenic nerve from WT (e) and Rag1 −/− (f) mice were electrically stimulated (STIM) or not (SHAMoperated), and norepinephrine (e) or acetylcholine (f) spleen contents were measured. (e, f) Data show mean ± S.E.M. of 2 independent experiments. (e,f) One-way ANOVA followed by Tukey's post hoc test were performed. *, p < 0.05; **, p < 0.01. splenic catecholaminergic fibers (Rosas-Ballina, 2008), which triggers the release of acetylcholine by CD4 + T cells (Rosas-Ballina, 2011) via β 2 adrenergic receptors (β 2-AR) (Vida, 2011). Acetylcholine then binds to the α7 nicotinic acetylcholine receptors (α7nAChR) on myeloid cells resulting in the inhibition of LPS-mediated production of pro-inflammatory cytokines (Olofsson, 2012). In apparent contrast to the mechanism described above, an anatomical and functional connection between the vagus and the splenic nerve could not be demonstrated by injecting anterograde tracers in the dorsal motor nucleus (DMV) of the vagus, and retrograde tracers in the spleen (Bratton, 2012;Cailotto, 2012). Furthermore, action potentials in the splenic nerve could not be detected in rats following vagal electrical stimulation (Bratton, 2012).
Whereas the presence of catecholaminergic fibers in the spleen is well established, the presence of cholinergic fibers has been debated for many years (Nance and Burns, 1989;Schäfer et al., 1998;Bellinger et al., 1993;Cano et al., 2001;Kooijman, 2015;Anderson et al., 2015), in part due to the various technical limitations that make the visualization of cholinergic structures by histochemical means challenging. While cholinergic markers have been widely used as an attempt to label cells, axons, and terminals in peripheral tissues, their lack of sensitivity has repeatedly been noted. Furthermore, immunostaining of cholinergic markers in spleen has been described as highly non-specific and variable by many investigators (Nance and Sanders, 2007;Stevens-Felten and Bellinger, 1997;Thayer and Sternberg, 2010). To overcome this problem, Gautron et al. used transgenic reporter mice in which the tdTomato fluorescent protein was selectively expressed in choline acetyltransferase (ChAT)-expressing cells, therefore allowing for the labeling of cholinergic neurons and their projections to the spleen (Gautron, 2013). Data showed the presence of few tdTomato-positive neuronal fibers around arterioles and not in association with the vasculature, diverging into lymphocytecontaining areas of the white pulp. However, this study may be questioned because the presence of cholinergic fibers revealed by this approach could have been due to the transient expression of the Cre (a, d) Inhibition of TNF production by nerve electrical stimulation in freely moving mice. WT (a), Foxn1 −/− (d) were implanted with electrodes onto the indicated nerves on day 0, and injected with LPS on day 7. Electrical stimulation was applied (STIM) or not (SHAM) and serum TNF levels were assessed in samples collected 90 min after LPS injection. Data show percent serum TNF levels in individual mice normalized to mean levels determined in SHAM control mice. (b, c) Differential effect of targeted nerve electrical stimulation on carotid blood pressure and heart rate in anesthetized mice before and after initiation of electrical stimulation. Representative heart rate pattern of an individual animal (b) and mean percent post-stimulation blood pressure (c, upper panel) and heart rate (c, lower panel) values of 6 mice. Baseline mean values were 432 ± 42.7 beats/min and 85 ± 8 mm Hg for heart rate and blood pressure respectively. Data show mean ± S.E.M. of 2-3 independent experiments (n ≥ 6 mice/group) in non-electrically stimulated (SHAM) and electrically stimulated (STIM) mice. (a, c, d) One-way ANOVA followed by Tukey's post hoc test were performed. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001. recombinase at an earlier stage of development of the neural fibers.
While some of the data described above may appear contradictory, it should be noted that several nerve-like structures project to the spleen in rodents. Unfortunately, most authors have referred to "the splenic nerve" in their publications without specifying the branch (or branches) that they were studying. Such an issue may be particularly problematic in experiments in which the function of the nerve fibers that project to the spleen was investigated by electrical stimulation or surgical ablation. This prompted us to perform an exhaustive anatomical and functional analysis of the nerves that project to the spleen in mice.

Results
Gross anatomy of spleen revealed three nerve-like structures among which two were associated with arteries as demonstrated by laser speckle imaging (Supp. Fig. 1, Fig. 1a). The two arterial nerves subdivided into two branches each and entered the spleen at non-apical locations (Fig. 1a). The third nerve referred to hereafter as the "apical splenic nerve" was not associated with an artery and entered the spleen at its apex (Fig. 1a). Both arterial and apical nerves were catecholaminergic as demonstrated by immunostaining. In contrast, the apical nerve but not the arterial nerves contained cholinergic fibers (Fig. 1b). Light sheet imaging of whole clarified spleen showed a dense network of catecholaminergic fibers across the entire spleen, and cholinergic fibers mainly located at the apex (Fig. 1c,d). In agreement with immunostaining data, stimulation of either the apical nerve or one of the two arterial nerves increased norepinephrine levels in spleen (5.82 ± 0.2 ng/mg unstimulated; 9.64 ± 0.76 ng/mg for vagus nerve stimulation; 10.3 ± 1.89 ng/mg for the arterial nerve; 10.9 ± 1.1 ng/ mg for the apical nerve) (Fig. 1e). In contrast, only apical nerve stimulation induced the release of acetylcholine spleen (21.8 ± 3.1 µg/ mg unstimulated; 38.8 ± 4.9 ng/mg for the arterial nerve; 73.4 ± 11.9 ng/mg for the apical nerve) in lymphocyte-deficient mice (Rag1 −/− ) ( Fig. 1f) further suggesting that cholinergic fibers, and not T cells, were the source of acetylcholine in these experiments.
We next investigated whether electrostimulation of the apical and arterial nerves could inhibit TNF production induced by lipopolysaccharide (LPS) injection in freely moving mice. We implanted mice with electrodes on intact apical, arterial or vagus (as a control) nerves, and applied electrostimulation one week later to stimulate both afferent and efferent fibers. We first checked that electrode implantation had no effect on LPS-induced TNF production (Supp. Fig. 3). All nerves were equally efficient at inhibiting LPS-induced TNF release when stimulated with the same electrical parameters (53.6 ± 8.5% for VN, 42.7 ± 6.7% for the arterial nerve, 49.2 ± 2.9% for the apical nerve) (Fig. 2a). However, and at variance with vagus nerve stimulation, neither arterial nor apical nerve stimulation caused changes in arterial blood pressure or heart rate ( Fig. 2b, c, Supp. Fig. 2). Therefore, both arterial and apical splenic nerve stimulation triggered an anti- inflammatory pathway without appreciable cardiovascular off-target effects.
Because the vagal anti-inflammatory pathway was shown to be dependent on CD4 + T cells (Rosas-Ballina, 2011), we investigated the role of these cells in the inhibition of inflammation by apical and arterial spleen nerve stimulation. As expected (Rosas-Ballina, 2011), the inhibition of LPS-induced TNF production by vagus nerve stimulation was abolished in T cell-deficient nude (Foxn1 −/− ) mice (Fig. 2d). This was also the case following arterial nerve stimulation (Fig. 2d). In striking contrast, the inhibition of LPS-induced TNF production by apical nerve stimulation was not abolished in Foxn1 −/− mice (3060 ± 380 pg/ml for SHAM vs. 1724 ± 192 pg/ml for the apical nerve) demonstrating that T cells were not required. Thus, the antiinflammatory effects of arterial and apical splenic nerve stimulation are mediated by distinct mechanisms. This prompted us to further investigate how apical nerve stimulation inhibited LPS-induced TNF secretion.
The inhibition of LPS-induced TNF production by apical nerve stimulation was abolished by both propranolol and methyllycaconitine (MLA), but not by atropine further suggesting that β 1 /β 2 -ARs and α7nAchRs, but not muscarinic AchRs were involved (Fig. 3a). Of note, a complete restoration of TNF production was achieved only when propranolol and MLA were administrated together suggesting the existence of two signaling pathways acting in parallel (Fig. 3a). TNF production was partially restored in Adrb2 −/− further confirming a role for β 2 -ARs in the anti-inflammatory effects of apical nerve stimulation (Fig. 3b).
Since the inhibition of LPS-induced TNF production by apical nerve stimulation did not require lymphocytes, we made the hypothesis that norepinephrine and acetylcholine acted directly on myeloid cells. To test this, we used LysM-Cre:Adrb2 fl/fl and LysM-Cre:Chrna7 fl/fl mice in which myeloid cells (Clausen et al., 1999) are selectively deficient in β 2-ARs and α7nAchRs respectively. While apical nerve stimulation did reduce LPS-induced TNF level in both LysM-Cre:Adrb2 fl/fl and Adrb2 fl/fl mice, LysM-Cre:Adrb2 fl/fl mice exhibited less TNF than Adrb2 fl/fl mice compared to Adrb2 fl/fl mice further demonstrating a role for β 2-ARs signaling in myeloid cells (81.6 ± 5.8% in LysM-Cre:Adrb2 fl/fl mice versus 60.9 ± 5.7% in Adrb2 fl/fl mice) (Fig. 3c). Furthermore, TNF production was partially restored in LysM-Cre:Chrna7 fl/fl further demonstrating a role for α7nAchR signaling in myeloid cells (67.1 ± 5.5% in LysM-Cre:Chrna7 fl/fl mice versus 48.6 ± 8.1% in Chrna7 fl/fl controls) (Fig. 3d). Lastly, TNF secretion was fully restored in LysM-Cre:Adrb2 fl/fl Chrna7 fl/fl mice in which myeloid cells lack both β 2-ARs and α 7-nAchRs (121.0 ± 17.8% in LysM-Cre:Adrb2 fl/fl Chrna7 fl/fl mice versus 60.5 ± 11.7% in Adrb2 fl/fl Chrna7 fl/fl controls) (Fig. 3e). Altogether, these results demonstrated that the inhibition of LPS-induced TNF secretion by apical nerve stimulation did not require the presence of lymphocytes and was dependent on β 2-ARs and α7nAchRs signaling in myeloid cells, with these two signaling pathways acting in parallel.
While the vagus anti-inflammatory pathway was first identified in rodents injected with LPS, vagus nerve stimulation (VNS) was eventually shown to ameliorates collagen-induced arthritis (CIA) development in rats (Levine, 2014). Promising effects have also been reported in patients with Rheumatoid arthritis (RA) (Koopman, 2016). Because apical nerve stimulation was as efficient as VNS at inhibiting LPS-induced TNF production, we investigated whether apical splenic nerve could also inhibit CIA in arthritis-prone DBA mice. Mice were immunized with collagen II on day 0, and implanted with electrodes applied onto the apical nerve on day 11. At day 21, animals were boosted with collagen II and followed for clinical symptoms until 35 days (Fig. 4a). Electrical stimulation was conducted every 4 h starting on day 16. Surgical implantation of electrodes onto the apical splenic nerve neither modified disease incidence nor progression compared to nonoperated animals (Supp. Fig. 4). While all SHAM mice developed arthritis within 28 days, disease onset was significantly delayed in stimulated animals, 16% of which showing complete protection throughout the study period (Fig. 4b). Apical nerve stimulation reduced disease severity (Fig. 4c, d) and joint swelling (Fig. 4e). This was accompanied by a reduction in synovial inflammation in the hind paw and reduction in tibiotalar and tarsus bone erosions in stimulated compared to SHAM mice. (Fig. 4f, g). Further, both the frequency and numbers of inflammatory monocytes were markedly reduced in the spleen of stimulated animals (Fig. 4h, i). Altogether, our results demonstrate that apical nerve stimulation inhibited CIA in mice.

Discussion
In this study, we have performed an exhaustive anatomical and functional analysis of the neural fibers that project to the spleen in mice. As expected, we found that the vast majority of the neural fibers that are present in the spleen are catecholaminergic as demonstrated by TH immunostaining. These fibers originated from three nerve branches among which two run along splenic arteries and one does not. This third nerve branch, which entered the spleen at its apex, has already been described but the nature of the fibers that it contains was not investigated (Buijs et al., 2008;Cailotto, 2012). While the origin of the cholinergic apical fibers is not known, they are likely to originate from either a vagus branch or the splanchnic (supra-renal ganglion). In agreement with immunostaining data, we found that electrical stimulation of any of these three nerve branches induced the release of norepinephrine in the spleen.
In addition to catecholaminergic fibers, we found that the apical nerve, but not the two arterial nerves, also contained cholinergic fibers as demonstrated by ChAT immunostaining. While some authors have already reported the presence of ChAT-positive neural fibers in the spleen, these findings have been debated for years in part due to the relatively low sensitivity and selectivity of ChAT immunostaining reagents and procedures. In agreement with our immunostaining data, we found that stimulation of the apical nerve, but not of the arterial nerves, increased acetylcholine spleen content in T lymphocyte-deficient mice, further suggesting that neural fibers, and not other acetylcholine producing cells, were the source of acetylcholine in these experiments.
Pioneer studies by Tracey's group has demonstrated the existence of a vagus anti-inflammatory reflex in which signals traveling in the vagus nerve modulate the activity of the splenic nerve, which secretes norepinephrine in spleen. In agreement with these data, stimulation of either the vagus nerve or one of the two arterial splenic nerve inhibited LPS-induced inflammation, and this phenomenon was abolished in T cell-deficient nude (Foxn1 −/− ) mice. In striking contrast, the inhibition of LPS-induced TNF production by apical nerve stimulation was neither abolished in Foxn1 −/− mice, nor in RAG-2 −/− mice that lack both T and B lymphocytes. Therefore, while apical and arterial splenic nerve stimulation were equally efficient at inhibiting LPS-induced secretion, they relied on different underlying mechanisms. We propose to refer to this newly described pathway as "apical splenic anti-inflammatory pathway", which co-exists with the previously described vagal antiinflammatory pathway. In contrast to the vagal anti-inflammatory pathway (Chavan et al., 2017), the apical splenic anti-inflammatory pathway does not require the presence of T cells and is dependent on β 2-ARs signaling in myeloid cells. While the apical splenic and the vagal anti-inflammatory pathways are both dependent on α7nAchRs signaling in myeloid cells, the former relies on the β 2-ARs and α7nAchRs pathways acting in parallel while the later relies on these pathways acting sequentially (Fig. 5). Another important difference between the vagal anti-inflammatory pathway and the new apical splenic anti-inflammatory pathway is that the former impacts a variety of organs (Martelli et al., 2019) including the gastrointestinal tract (Matteoli and Boeckxstaens, 2013) and the kidneys (Inoue, 2016) while the latter targets the spleen only. It remains to be determined whether other vagus-unrelated peripheral nerves could convey an anti-inflammatory signal to specific tissues, and more specifically to organ-draining lymph nodes. RA is a chronic degenerative autoimmune disease characterized by joint synovial inflammation and bone cartilage erosion leading to significant disabilities (Aletaha and Smolen, 2018). While symptomatic relief can be achieved by treatment with anti-TNF antibodies and other biologicals, a sizeable proportion of RA patients do not respond to these treatments (Aletaha and Smolen, 2018). Consistent with the critical role of TNF in the pathogenesis of RA (Vervoordeldonk and Tak, 2002), vagus nerve stimulation ameliorates collagen-induced arthritis (CIA) development in rats (Levine, 2014) and promising effects have been reported in early stage clinical studies in RA patients (Koopman, 2016). While VNS is currently approved for the treatment of drug resistant epilepsy and depression with no major side-effects, it sometimes carries the risks of untoward off-target effects depending on the electrostimulation parameters and the sensitivity of the patient (Ben-Menachem, 2001). VNS produced heart rate reductions in most preclinical species involving rodents as well as large animals (Warner and Russell, 1969;Buschman, 2006). However electrostimulation of apical splenic nerve is not associated with significant impact on cardiovascular parameters in mice in accordance with the absence of detectable afferent fibers in this nerve. Thus, at variance with vagus nerve stimulation, apical nerve stimulation may offer the opportunity to deliver a boarder panel of electrical parameters of stimulation that may improve therapeutic efficacy. While it remains to be determined whether apical splenic nerve stimulation is as efficient as VNS at inhibiting CIA, stimulating the splenic nerve instead of the vagus nerve might offers an alternative for RA patients. On another topic, the mechanisms underlying the beneficial impact of apical splenic nerve stimulation on CIA remain to be identified. Based on our data in the LPS model, we could hypothesize that it involves the β 2-ARs and α7nAchRs pathways acting in parallel in myeloid cells to inhibit the secretion of pro-inflammatory cytokines. Whatever the case, other mechanisms may also be at play including the mobilization of regulatory T cells.
These results pave the way for the use of splenic nerve electrical stimulation as a promising alternative to other therapeutic strategies in RA with less side effects.

Acknowledgements
This work was funded by a phase 2 collaborative research grant from Galvani Bioelectronics. This work was also supported by the LABEX SIGNALIFE (#ANR-11-LABX-0028-01) and the FHU OncoAge.

Competing financial interest
A patent application is pending.

Methods
Mice: C57BL/6, DBA, Foxn1 -/-, Rag1 -/-, LysM-Cre:Adrb2 fl/fl and LysM-Cre:Chrna7 fl/fl mice were purchased from Charles River. ADRB2 ko , mice were purchased from The Jackson Laboratory and backcrossed on the C57BL/6 background for at least 10 generations. All experiments were Fig. 5. Schematic representation of the anti-inflammatory pathways at play in the spleen. The upper part of the scheme represents the mechanism of vagal anti-inflammatory pathway relying exclusively on the production of norepinephrine by splenic catecholaminergic fibers (red), which triggers the release of acetylcholine by CD4 + T cells via β2 adrenergic receptors (β2-AR). Acetylcholine then binds to the α7 nicotinic acetylcholine receptors (α7nAChR) on myeloid cells inhibiting LPS-mediated inflammatory cytokine production. The lower part of the scheme represents the mechanism of apical splenic anti-inflammatory pathway, which relies the production of norepinephrine and acetylcholine by splenic catecholaminergic (red) and cholinergic (green) fibers respectively. These neurotransmitters then bind to β2-ARs and α7nAchRs expressed by myeloid cells, with these two signaling pathways acting in parallel to inhibit LPS-mediated inflammatory cytokine production. performed with 8-12 wk old female mice except for the DBA strain which were 8 wk old male mice. Mice were housed on a 12 h light/dark cycle (lights on/off at 7 am/7 pm) with food ad libitum. Mice were treated in accordance with our local Animal Care and Use Committee guidelines.
Laser speckle imaging: After anesthesia, the spleen tissue was exposed and placed 30 cm below the Moor-FLPI laser speckle perfusion imager (Moor instruments Ltd.). Blood perfusion images were saved and analyzed by the Image Review software (Moor-FLPI-V2.0).
Immunohistochemistry: Nerves were excised and immediately submerged in OCT compound (Sakura Finetek, Torrance, CA). Tissues in OCT were quickly frozen using dry ice, then kept at -80°C for long-term storage. Nerve cryosections were cut in triplicate at 6 μm using a Microm HM 550 cryostat (Thermo Fisher Scientific, Inc.). Sections were fixed with cold acetone/methanol (v/v: 1: 1) for 5 min, then washed in PBS. Primary antibodies (anti-choline acetyl transferase, AB144P, Merck; anti-tyrosine hydroxylase, AB152, Merck; anti-neurofilament, ab4680, Abcam) were diluted 1:50, 1:300 and 1:500 respectively in PBS and incubated for 1 hour in a humidified chamber. Sections were washed twice with PBS, and secondary antibodies were applied at 1:400 dilutions for 45 min (Donkey anti-goat, Donkey Anti-Rabbit, Donkey Anti-Chicken IgY resp. from Jackson Immunoresearch). Slides were washed twice in PBS for 5 min, fixed with 4% paraformaldehyde, treated with 1% glycerol in PBS before coverslipping with ProLong Gold Antifade Mountant (Thermo Fisher Scientific, Inc.) and imaged under a fluorescent microscope. Images were analyzed with Volocity image analysis software (PerkinElmer, Waltham, MA).
Light-sheet-based fluorescent microscopy: Animals were perfused using a 4% PFA solution followed by a PBS solution. Excised spleens were depigmented and clarified using the iDisco+ method (https://idisco.info/). Briefly, they were dehydrated at room temperature in successive bathes of 20% MetOH for 1h, 40% MetOH for 1h, 60% MetOH for 1h, 80% MetOH for 1h, 100% MetOH for 1h and 100% MetOH overnight. Then the organs are incubated in a solution of 33% MetOH and 66% Di-ChloroMethan (DCM, Sigma) overnight and washed twice with 100% methanol for 1h. Organs were then bleached in chilled fresh 5% H 2 0 2 in methanol overnight at 4°C before being rehydrate with methanol/H2O series (80%, 60%, 40%, 20% and PBS, 1 h each at RT). Samples were them immunolabelled for 24 h with anti-TH (AB152, Merck) and anti-ChAT (AB144P, Merck) after overnight permeabilization at 37°C and overnight blocking in blocking solution (1,7% TritonX-100, 6% donkey serum and 10% DMSO in PBS). After 3 washes in PBS/ 0.2%Tween-20, samples were incubated with the secondary antibody (donkey anti-goat, donkey anti-rabbit resp. from Jackson Immunoresearch) for 24h. Samples were then dehydrated at room temperature in successive bathes of 20% MetOH for 1h, 40% MetOH for 1h, 60% MetOH for 1h, 80% MetOH for 1h, 100% MetOH for 1h and 100% MetOH overnight. Then the organs are incubated in a solution of 33% MetOH and 66% Di-ChloroMethan (DCM, Sigma) for 3h at RT, then in 100% DCM twice for 15 min twice and transferred overnight into the clearing medium 100% DiBenzylEther 98% (DBE, Sigma). Imaging was performed using a home-made light-sheet ultramacroscope. The specimen was placed into a cubic cuvette filed with DBE placed on the Z-stage of the bench. It was illuminated with planar sheets of light, formed by cylinder lenses. The light coming from a multi-wavelength (561 nm) laser bench (LBX-4C, Oxxius) was coupled via two single mode optical fibers into the setup, allowing illumination from one or two sides. We used two-sided illumination. The specimen was imaged from above with a MVX10 macroscope, through a PlanApo 2X/0.5 NA objective (Olympus) with an additional zoom of the macroscope of 1.6, which was oriented perpendicular to the 561nm light sheet. Images were captured using a sCMOS Camera (Orca-Flash4.0) synchronized with the z-stage moving the sample through the light sheet. The ultramicroscope is managed by Micro-manager software and z-stacks of images were taken every 2 µm. The images stacks are fused using the alpha-blending method with a home-made ImageJ macro (Rasband, W.S., ImageJ, U. S. National Institutes of Health, Bethesda, Maryland, USA, http://imagej.nih.gov/ij/, 1997e2012).
Electrodes and surgery: For studies in anaesthetized animals ( Fig. 1e, g, h), mice were pre-medicated with buprenorphine (100 µg/kg, i.p.) 30 min before surgery and anaesthetized with isoflurane (2% v/v). A hook electrode was placed under the splenic or the vagus nerve. For studies in freely moving animals, mice were pre-medicated with buprenorphine (100 µg/kg, i.p.) 30 min before surgery and anaesthetized with isoflurane (2% v/v) for the duration of the surgery. For splenic nerve implantation, one mm length 100 µm-sling bipolar micro-cuff electrodes (CorTec) were implanted onto the arterial or apical nerve. For vagus nerve implantation, two mm length 200 µm-tunnel bipolar micro-cuff electrodes (CorTec) were implanted onto the vagus nerve.
Electrostimulation: Mice were anesthetized and either 100 µm sling or 200 µm tunnel Cortec electrodes were implanted onto the splenic nerves (apical or arterial) or the vagus nerve respectively. Seven days following surgery, these animals were injected ip with a sublethal dose of LPS (5 mg/kg) which has been used in previous studies to induce LPS production (Huston, 2006). Electrostimulation was applied using a PlexStim V2.3 (Plexon) starting at −10, 0 and +20 min relative to LPS injection. Sera was collected at 90 min after LPS injection and assessed for TNF levels. Controls consist of fully Cortec implanted mice, which did not receive electrical stimulation (SHAM). Electrostimulation were rectangular charged-balanced biphasic pulses with 650 µA pulse amplitude, 100 µs pulse width (positive and negative) at 10 Hz frequency for 2 min (STIM).
Carotid blood pressure: C57/BL6 mice were anaesthetized as described above, a catheter was placed into the left carotid, connected to a pressure transducer and recorded using a Acqknowledge 881 (MP100WS) software (Biopac System, Inc). A tunnel or cuff electrode were placed under the vagus nerve or the splenic respectively and different patent of stimulation were delivered starting by the lowest instensity. Mean blood pressure and heart rate were analyses using Biopac data acquisition Acqknowledge 881 (MP100WS) software (Biopac System, Inc).
TNF, norepinephrine and acetylcholine levels: For TNF, retro-orbital blood sampling was performed under isofluorane anesthesia. TNF levels were measured by ELISA (Mouse TNF-alpha DuoSet, R&D Systems) following manufacturer instructions and normalized to those measured in SHAM animals. TNF was below the lower level of detection (LLOD) in control mice that were not injected with LPS. For acetylcholine, mice received 0.1 µg/g of the acetylcholine esterase inhibitor neostigmine intraperitoneally 30 mn prior to electrostimulation. Spleen were harvested and snap-frozen in liquid nitrogen immediately after electrostimulation. Protein extracts were prepared using Precellys machine (MP Biomedicals) in Lysing Matrix tube with TNET Buffer (10 mM Tris 150 mM NaCl 5 mM EDTA 1% Triton 10% Glycerol 1%, Protease Phosphatase Inhibitor Cocktail (Pierce, Thermo Fisher Scientific)). Protein concentration were assessed using BCA Protein Assay KIT (Pierce, Thermo Fisher Scientific) following manufacturer instructions. Norepinephrine or acetylcholine levels were measured by ELISA (for norepinephrine: Norepinephrine -Sensitive, DLD Diagnostika GmbH; for acetylcholine: QuickDetect™, Clinisciences) following manufacturer recommendations. LLOD were 15, 75 and 1.2 pg/ml for TNF, norepinephrine and acetylcholine respectively. Treatment with antagonists: Antagonists (atropine: 1 mg/kg (Pinardi et al., 2003); Proparnolol: 5 mg/kg (Abdin et al., 2014); methyllycaconitine: 5 mg/kg (Lewis et al., 2015)) were given i.p. 30 min prior to LPS injection to allow for the antagonists to inhibit all accessible receptors (Albanus et al., 1968;Hanson et al., 1978;Stegelmeier et al., 2003).
Induction and assessment of collagen-induced arthritis: Bovine type II collagen (2 mg/ml in 0.05 M acetic acid; Chondrex, Redmond, WA) was mixed in an equal volume of Freund's complete adjuvant (2 mg/ml of Mycobacterium tuberculosis; Chondrex). DBA mice were immunized intradermally at the base of the tail with 100 μl of emulsion (100 μg collagen) on day 0 and day 21. At day 11, mice were anaesthetized with isoflurane and the spleen area was exposed. One mm length 100 µmsling micro-cuff electrode (CorTec) was implanted onto the apical splenic nerve. At day 16, mice were placed in individual cage and connected to a PlexStim V2.3 (Plexon) or MAPS (Axonic) stimulator. The set-up of the electrostimulation were rectangular charged-balanced biphasic pulses with 650 µA pulse amplitude, 2 ms pulse width (positive and negative) at 10 Hz frequency for 2 min 6 times a day (every 4 hours). The severity of arthritis was assessed using an established semiquantitative scoring system of 0-4, where 0 = normal, 1 = swelling in 1 joint, 2 = swelling in > 1 joint, 3 = swelling in the entire paw, and 4 = deformity and/or ankylosis. The cumulative score for all 4 paws of each mouse (maximum possible score 16) was used to represent overall disease severity and progression. For the evaluation of incidence, mice were considered to have arthritis if the clinical arthritis score was at least at 1 point for three consecutive days.
Statistics: CIA progression was plotted using Kaplan-Meier's curves and differences between groups were estimated using the log-rank test. Normality of sample distribution was assessed using the Kolmogorov-Smirnov test. For comparison between two groups, statistical significance was assessed using unpaired t-test. For comparison between more than two groups, statistical significance was assessed using oneway ANOVA followed by Tukey's post hoc test. All statistical analysis were performed using GraphPad Prism v.6.